Search Results for author:
Found 11 papers, 6 papers with code
In-n-Out: Calibrating Graph Neural Networks for Link Prediction
While networks for tabular or image data are usually overconfident, recent works have shown that graph neural networks (GNNs) show the opposite behavior for node-level classification.
Human-in-the-Loop Causal Discovery under Latent Confounding using Ancestral GFlowNets
Surprisingly, while CD is a human-centered affair, no works have focused on building methods that both 1) output uncertainty estimates that can be verified by experts and 2) interact with those experts to iteratively refine CD.
Locking and Quacking: Stacking Bayesian model predictions by log-pooling and superposition
Currently, these predictive distributions are almost exclusively combined using linear mixtures such as Bayesian model averaging, Bayesian stacking, and mixture of experts.
Distill n' Explain: explaining graph neural networks using simple surrogates
We also propose FastDnX, a faster version of DnX that leverages the linear decomposition of our surrogate model.
Provably expressive temporal graph networks
Specifically, novel constructions reveal the inadequacy of MP-TGNs and WA-TGNs, proving that neither category subsumes the other.
Parallel MCMC Without Embarrassing Failures
While efficient, this framework is very sensitive to the quality of subposterior sampling.
Online graph nets
To avoid temporal message passing, OGN maintains a summary of the temporal neighbors of each node in a latent variable and updates it as events unroll, in an online fashion.
Rethinking pooling in graph neural networks
In this paper, we build upon representative GNNs and introduce variants that challenge the need for locality-preserving representations, either using randomization or clustering on the complement graph.
Federated Stochastic Gradient Langevin Dynamics
We apply conducive gradients to distributed stochastic gradient Langevin dynamics (DSGLD) and call the resulting method federated stochastic gradient Langevin dynamics (FSGLD).
Learning GPLVM with arbitrary kernels using the unscented transformation
Gaussian Process Latent Variable Model (GPLVM) is a flexible framework to handle uncertain inputs in Gaussian Processes (GPs) and incorporate GPs as components of larger graphical models.
Embarrassingly parallel MCMC using deep invertible transformations
While MCMC methods have become a main work-horse for Bayesian inference, scaling them to large distributed datasets is still a challenge.
